\bigskip
\newcolumntype{C}{>{\centering\arraybackslash}X}

\caption*{Panel B: Lee Bounds}
\begin{tabularx}{\linewidth}{lCCCCCCCCCC}

\midrule
&{(1)}&{(2)}&{(3)}&{(4)}&{(5)}&{(6)}&{(7)}&{(8)}&{(9)}&{(10)} \tabularnewline \midrule


HEE&0.059***&0.079***&0.025&0.046**&0.094***&0.125***&0.037**&0.060***&0.135***&0.166*** \tabularnewline
&(0.014)&(0.014)&(0.018)&(0.019)&(0.026)&(0.028)&(0.016)&(0.016)&(0.033)&(0.034) \tabularnewline
HEEC&0.044***&0.068***&--0.008&0.017&0.048*&0.084***&0.020&0.052***&0.162***&0.199*** \tabularnewline
&(0.014)&(0.014)&(0.019)&(0.019)&(0.025)&(0.026)&(0.016)&(0.017)&(0.031)&(0.032) \tabularnewline
\bottomrule 

\end{tabularx}
\captionsetup{justification=justified, width=\linewidth} \caption*{\footnotesize \textit{Notes:} This table presents bounds on the treatment effects estimated in Table \ref{table-endline-knowledge-aggregate}. Panel A uses extreme value (Manski) bounds following \textcite{horowitz2000nonparametric}. Panel B uses trimming bounds following \textcite{lee2009training} and is estimated with no covariates.} \end{table}
